Lemma 2.2 (Minty), p. 3 — for continuous monotone F, solutions of (1.1) are the solutions of the dual VI
ProvedProjReflGrad.Weak.lemma_2_2mintymonotone-operatorp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1variational-inequality
Let be a real inner product space, convex, and continuous on and monotone on , i.e. for all . Then is a solution of (1.1), that is and for all , if and only if solves
The dual formulation is linear in , which makes it stable under weak limits; it is how weak cluster points of the iterates are shown to be solutions.
Formalization Note. The paper takes ; here is defined on all of and only its restriction to is constrained. Convexity of is the standing assumption of (1.1). Closedness and completeness are not needed and not assumed.
Preamble
import Mathlib import Definitions.Def_ProjReflGrad_Weak_Setting
Formal statement
namespace ProjReflGrad.Weak
/-- Lemma 2.2 (Minty) (Malitsky 2015, p. 3): for `C` convex and `F` continuous and monotone on `C`,
`x*` solves (1.1) iff `x* ∈ C` and `⟨F(y), y - x*⟩ ≥ 0` for all `y ∈ C`. -/
theorem lemma_2_2 {H : Type*} [NormedAddCommGroup H] [InnerProductSpace ℝ H]
(C : Set H) (hCcv : Convex ℝ C) (F : H → H) (hcont : ContinuousOn F C)
(hmono : ∀ x ∈ C, ∀ y ∈ C, 0 ≤ inner ℝ (F x - F y) (x - y)) (xs : H) :
xs ∈ solSet C F ↔ xs ∈ C ∧ ∀ y ∈ C, 0 ≤ inner ℝ (F y) (y - xs) := by sorry
end ProjReflGrad.Weak
Source
Malitsky, Projected Reflected Gradient Methods for Monotone Variational Inequalities, arXiv:1502.04968v1, p. 3, Lemma 2.2 (Minty)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.